[SOLVED] CCITT psophometric filter ckt

[Orson Cart]

Hello all,
Can some one direct me to or supply a schematic that gives the psophometric filtering function, i.e. 0dB at 800Hz with a pass band that falls sharply near 200Hz and 4kHz?

all pointers gratefully received

 

[FvM]

The filter characteristic is defined in recommendation O.41 https://www.itu.int/rec/T-REC-O.41-198811-S . I would fit an active or LC filter to it using parameter optimization.

 

[Orson Cart]

Thanks, I have this info, was really wanting a circuit (analog) as per the posting.

 

[FvM]

Yes, I know. For lack of a ready-designed circuit I suggested a method to calculate it.

I also found the HP 3556 service manual published at agilent.com. Unfortunately, the schematics aren't readable

 

[Orson Cart]

Correct the schematics are completely unreadable....

 

[FvM]

I tried a filter based on real poles (first order filters, similar to the well-known sound pressure level weighting functions). Together with a Butterworth filter for the steep cut-off, I was able to stay at least within the defined mask. For a better fit, you'll probably need complex poles for lower frequencies as well.

 

[Orson Cart]

Looks very good, are we any closer to a schematic with op-amps, R's, C's... ?

 

[FvM]

 

[Orson Cart]

Very good FVM, many thanks...!

 

[CJS]

I would like to make a Matlab emulation of the psophometer filter as an IIR filter. Has anyone on this thread reduced the op-amp circuit to s-plane or z-plane complex poles and zeros or equivalent transfer function form?

 

[FvM]

The annotation of the curve in post #6 translates directly to a s-domain description (if you know the definition of he butterworth polynomials). https://en.wikipedia.org/wiki/Butterworth_filter

It's also known by Matlab, I think.

 

[Audioguru]

The filter cuts off all the important consonant sounds of speech but passes grunts and groans (vowels). Just like telephones sounded for the last 134 years.